Chern and Pontryagin classes of manifold constructed as compactification of vector bundle

89 Views Asked by At

I'm trying to compute the total Chern class of the tangent bundle of a manifold constructed as a "compactification" or "projective completion", but I'm not sure I quite have all the details under control.

The manifold $M_{8}$ is eight-dimensional and complex, and constructed as a two-sphere bundle over a six-dimensional (Kahler) base space $B_{6}$. I want to think of this bundle as the total space of the anti-canonical line bundle $\mathcal{L}$ of $B_{6}$, together with a point added at infinity in each $\mathbb{C}$ fibre to make them into spheres. In practice, from here, I believe this means I can think of $M_{8}$ as the total space of $\mathbb{P}(\mathcal{O}\oplus\mathcal{L})$, where $\mathbb{P}$ indicates the projectivisation and $\mathcal{O}$ is the trivial line bundle over $B_{6}$.

I then want to compute the total Chern class of $M_{8}$ in terms of the data of $B_{6}$. As a start, following this, I have that, taking $V=\mathcal{O}\oplus\mathcal{L}$ and $\pi\colon\mathbb{P} V\to B_{6}$, $$ c(T_{\mathcal{M}_{8}})=c(\pi^{*}(\mathcal{O}\oplus\mathcal{L})\otimes\gamma^{*})c(\pi^{*}T_{B_{6}}), $$ where $\gamma$ is the tautological line bundle over $\mathbb{P} V=\mathcal{M}_{8}$. Since $\mathcal{L}$ is the anti-canonical bundle of $B_{6}$, I also know that $c_{1}(\mathcal{L})=c_{1}(T_{B_{6}})$. The Whitney sum formula, etc., then gives \begin{align*} c(T_{\mathcal{M}_{8}}) & =c(\pi^{*}\mathcal{O}\otimes\gamma^{*})c(\pi^{*}\mathcal{L}\otimes\gamma^{*})c(\pi^{*}T_{B_{6}})\\ & =\left(1+\pi^{*}c_{1}(\mathcal{O})+c_{1}(\gamma^{*})\right)\left(1+\pi^{*}c_{1}(T_{B_{6}})+c_{1}(\gamma^{*})\right)\pi^{*}c(T_{B_{6}}), \end{align*} which I can expand out to read off the Chern classes. At this point, my geometry runs out. I have two questions:

  1. Naively, I'd say that $\gamma^{*}=\pi^{*}\mathcal{O}(1)$ and would then need to know $\mathcal{L}$ in terms of $\mathcal{O}(n)$ in order to make progress. Does that sound correct, or is that too quick? From this, it seems that it might be more complicated...
  2. In fact, the end goal is to compute the Pontryagin classes for $\mathcal{M}_{8}$. In particular, is it obvious that $p_{1}^{2}$ and $p_{2}$ are always trivial, since $\mathcal{M}_{8}$ is constructed as a bundle? (Since everything reduces to classes $B_{6}$, there are no non-trivial eight-form classes constructed from the Chern classes?)

Thank you in advance, and sorry if I've said something imprecise. I think this is fairly standard material, but I haven't come across it before.

Edit: I think the point is to use the tautological exact sequence $$ 0 \to \gamma \to \pi^* V \to T_{\mathbb{P}(V)/B_6}\otimes \gamma \to 0 $$ and then the observation that the Chern classes of $V$ satisfy (Bott & Tu, page 270) $$ x^n + \pi^* c_1(V) x^{n-1} + \dots + \pi^* c_n(V) =0 $$ where $x = c_1(\gamma^*)$ and $V$ is rank $n$. For $V=\mathcal{O} \oplus \mathcal{L}$, we then have $$ x^2 + \pi^* c_1(\mathcal{L})x=0 $$ so that $x = c_1(\gamma^*) = - \pi^* c_1(\mathcal{L})$. (In particular, $\gamma^* \neq \pi^*\mathcal{O}(1)$, but instead picks up a dependence on $\mathcal{L}$.) With this in hand, one can use the above formula for $c(T_{\mathcal{M}_{8}})$ to compute the Chern classes.